Metagenome-based behcet disease uveitis biomarker combination as well as screening method and application thereof

Abstract

The invention relates to the field of biological medicine, in particular to a BU (Behcet's disease uveitis) biomarker screening method based on an intestinal microorganism metagenome, a screened BU biomarker combination and application of the biomarker combination in BU diagnosis. The kit is especially suitable for early diagnosis of BU and differential diagnosis of other systemic immune diseases. The biomarker combination consists of 24 intestinal strains, and is obtained by analyzing excrement metagenome data of BU patients, other systemic immune disease patients and healthy controls through diversity analysis, LEfSe screening and ROC verification. The marker has high sensitivity and specificity, can effectively distinguish BU from healthy people and diseases such as VKH, AS, RA, CD and UC, is suitable for developing a non-invasive BU diagnostic kit, and remarkably improves the clinical diagnosis efficiency.

Description

Technical Field

[0001] This invention relates to the field of biomedicine, specifically to a method for screening biomarkers for Behçet's disease uveitis (BU) based on gut microbial metagenomics, a combination of screened Behçet's disease uveitis biomarkers, and the application of this biomarker combination in the diagnosis of BU, particularly suitable for the early diagnosis of BU and the differential diagnosis of BU with other systemic immune diseases. Background Technology

[0002] Behçet's disease is a systemic, self-inflammatory disease characterized primarily by vasculitis. It predominantly affects young adults and is characterized by multi-system involvement, typically manifesting as recurrent oral ulcers, genital ulcers, skin lesions, and ocular inflammation. Ocular involvement occurs in 50%-70% of patients, usually presenting as Behçet's uveitis (BU). BU primarily affects intraocular structures such as the iris, ciliary body, and choroid. Its inflammatory process is characterized by recurrent attacks and persistent inflammation; approximately 25%-45% of patients experience irreversible ocular tissue damage due to repeated inflammation, ultimately leading to blindness. The chronic progression and high disabling nature of BU not only cause visual impairment and a decline in quality of life but also impose a significant socioeconomic burden due to its primary impact on the working-age population.

[0003] Currently, the diagnosis of BU relies primarily on the patient's clinical presentation. However, this diagnostic model has significant limitations:

[0004] First, due to the heterogeneity of clinical manifestations, early clinical symptoms are often atypical and easily confused with other types of uveitis and systemic immune diseases. For younger physicians, misdiagnosis and missed diagnosis are common, leading to delays in treatment.

[0005] Secondly, in terms of disease progression, BU often recurs, and each recurrence carries the risk of irreversible vision loss and permanent eye complications;

[0006] Furthermore, by the time a patient presents with typical eye symptoms, irreversible vision damage may have already occurred.

[0007] Therefore, in addition to conventional diagnostic methods, reliable and non-invasive auxiliary examination methods are of great significance for the early identification, timely treatment, prevention or delay of the occurrence of severe symptoms in patients with bronchiolitis (BU) and for reducing long-term disability rates.

[0008] In recent years, the role of gut microbiota dysbiosis in the pathogenesis of ulcerative colitis (BU) has received increasing attention. Multiple studies have shown that gut microbiota dysbiosis, along with other diseases, is closely associated with various immune-mediated disorders, including Vogt-Koyanagi-Harada syndrome (VKH), ankylosing spondylitis (AS), rheumatoid arthritis (RA), Crohn's disease (CD), and ulcerative colitis (UC). Significant differences in gut microbiota composition between BU patients and healthy individuals have also been observed. Animal experiments have confirmed that transplanting fecal microbiota from BU patients into mice resulted in abnormal activation of the IL-23 / IL-17 signaling pathway and exacerbated the severity of experimental autoimmune uveitis (EAU). Furthermore, the gut microbiota provides a source of cross-reactive antigenic peptides, which may contribute to BU pathogenesis by triggering molecular mimicry and activating autoreactive T cells.

[0009] To improve the diagnostic accuracy of early Behçet's disease, many technicians have developed various technical solutions. For example, CN108290058A discloses a method for stratifying and identifying patients with IL23-mediated disease who respond to treatment with IL23 antagonists by using components of the IL23 pathway, including IL22, LCN2, and combinations thereof, as biomarkers. The method uses IL23 pathway biomarker levels above or below a predetermined threshold to determine whether the patient has IL23-mediated disease or symptoms that qualify for treatment. However, technicians have found this technical solution to be extremely cumbersome in practice, resulting in a complex testing process and limited practicality.

[0010] In other words, the traditional methods of diagnosing Behçet's disease in the early years had at least the following shortcomings:

[0011] 1. The diagnosis of Behçet's disease mainly relies on the patient's clinical manifestations. There is a lack of reliable and non-invasive auxiliary examination methods, which leads to low diagnostic efficiency, high misdiagnosis rate, and easy problems such as atypical early symptoms and confusion with other uveitis and systemic immune diseases.

[0012] 2. The detection methods are complex and cumbersome to operate, making it difficult to achieve rapid and efficient diagnosis. Furthermore, the detection results are easily affected by platform and population differences, which affects the accuracy and repeatability of the detection.

[0013] 3. The lack of specific microbial markers that can simultaneously distinguish Behçet's disease from other immune-mediated diseases makes it difficult to achieve accurate diagnosis and differential diagnosis of different diseases;

[0014] 4. When detecting microbial structure or functional disorders, there is a lack of comprehensive analytical capabilities regarding the composition and function of gut microbiota, making it difficult to accurately assess the relationship between disease and health status.

[0015] At the end of 2021, technicians developed some new technical solutions in an attempt to address the aforementioned problems. Examples include publications such as "Behçet's Disease Marker Genes and Their Applications" (CN114182007A) and "Behçet's Disease Marker Microorganisms and Their Applications" (CN114317784A). However, in practical applications over the past few years, technicians have discovered that these technologies still have significant problems:

[0016] (1) The marker gene disclosed in CN114182007A targets Behçet's disease (BD) as a whole, without distinguishing whether ocular involvement (i.e., Behçet's disease uveitis, BU) exists in BD patients. BU is one of the most severe subtypes of BD (50%-70% of patients have ocular involvement, 25%-45% experience blindness), and its gut microbiota characteristics may differ from those of BD patients without ocular involvement. This approach cannot accurately identify BU patients, potentially leading to delays in early diagnosis and intervention of BU.

[0017] Furthermore, the 13 core genetic markers disclosed in CN114182007A are based solely on a comparison between BD and healthy individuals, and do not include metagenomic data from other diseases. This may lead to the omission of unique gut microbiota characteristics specific to BU and distinguishing them from other diseases. Therefore, the diagnostic efficacy of CN114182007A is limited to differentiating between BD and healthy individuals, and cannot meet the complex clinical needs for differentiating BU.

[0018] Furthermore, the genetic markers disclosed in CN114182007A are only used to distinguish BD patients from healthy individuals and do not address the differentiation from other systemic immune diseases with similar clinical manifestations (such as Vogt-Koyanagi-Harada syndrome (VKH) and ankylosing spondylitis (AS)). Since early symptoms of BU are often confused with uveitis such as VKH or immune diseases such as AS, using this technical approach could easily lead to misdiagnosis.

[0019] (2) The microbial markers in CN114317784A target BD as a whole and do not consider BU-specific ocular inflammatory mechanisms (such as abnormal activation of the IL-23 / IL-17 pathway). Studies have shown that gut microbiota dysbiosis in BU can trigger ocular autoimmune responses through cross-reactive antigenic peptides, but the seven microorganisms disclosed in CN114317784A (such as Atkinsonella texensis) are not associated with BU-specific immune pathways and may not reflect the core pathological features of BU.

[0020] The microbial markers in CN114317784A are primarily used to differentiate BD patients from healthy individuals; their ability to differentiate them from other types of uveitis (such as VKH) or systemic immune diseases has not been verified. Clinically, both BU and VKH present as uveitis with similar symptoms; using the diagnostic method disclosed in CN114317784A may lead to confusion between the two.

[0021] The biomarker in CN114317784A was only compared with BD and healthy individuals, and its AUC in the validation set was only 89.8%. However, the clinical diagnosis of BU requires the simultaneous exclusion of multiple similar diseases. The biomarker disclosed in CN114317784A may have a false positive risk in the context of multiple diseases, and its diagnostic efficacy is not high.

[0022] In summary, current diagnostic methods for Behcet's disease employ markers with limited coverage and narrow identification range. Therefore, these diagnostic methods still suffer from insufficient specificity and cannot be effectively focused on the diagnosis of BU. Summary of the Invention

[0023] The purpose of this invention is to address the shortcomings of existing technologies by providing a combination of biomarkers for Behçet's disease and uveitis (BU) based on a combined metagenomic model of multiple diseases, along with its screening method and applications. The biomarker combination of this invention consists of 24 gut microbiota species, obtained through analysis of fecal metagenomic data from BU patients, patients with other systemic immune diseases, and healthy controls, followed by diversity analysis, LEfSe screening, and ROC validation. This biomarker exhibits high sensitivity and specificity, effectively distinguishing BU from healthy individuals and diseases such as VKH, AS, RA, CD, and UC. It is suitable for developing non-invasive BU diagnostic kits, significantly improving clinical diagnostic efficiency.

[0024] The objective of this invention is achieved through the following approach:

[0025] A biomarker set for the diagnosis of Behçet's disease uveitis, comprising microorganisms from the genera *Bilophila*, *Alistipes*, *Dorea*, *Blautia*, *Coprococcus*, *Lachnospiraceae*, *Bacteroides*, *Paraprevotella*, *Streptococcus*, *Eubacterium*, *Oscillibacter*, *Phascolarctobacterium*, *Burkholderiales noname*, *Parasutterella*, *Enterobacter*, and *Escherichia*.

[0026] Preferably, the biomarker combination includes the following 24 bacterial species:

[0027] Bilophila, a member of the genus Bilophila;

[0028] Alistipes putredinis, Alistipes shahii, Alistipes sp AP11 of the genus Alistipes;

[0029] Dorea formicigenerans, Dorea longicatena of the genus Dorea;

[0030] Ruminococcus gnavus, Ruminococcus obeum, Ruminococcus torques of the genus Blautia;

[0031] Coprococcus comes of the genus Coprococcus;

[0032] Lachnospiraceae bacterium 5_1_63FAA of the family Lachnospiraceae;

[0033] Bacteroides coprophilus, Bacteroides eggerthii, Bacteroides plebeius of the genus Bacteroides;

[0034] Paraprevotella of the genus Paraprevotella;

[0035] Streptococcus salivarius, Streptococcus thermophilus of the genus Streptococcus;

[0036] Eubacterium hallii of the genus Eubacterium;

[0037] Oscillibacter unclassified of the genus Oscillibacter;

[0038] Phascolarctobacterium succinatutens of the genus Phascolarctobacterium;

[0039] Burkholderiales bacterium 1_1_47 of the genus Burkholderiales noname;

[0040] Parasutterella excrementihominis of the genus Parasutterella;

[0041] Enterobacter cloacae of the genus Enterobacter;

[0042] Escherichia is a species of Escherichia.

[0043] The present invention also includes a method for screening microorganisms in the above-mentioned combination of biomarkers, comprising the following steps:

[0044] 1) Obtain metagenomic sequencing data and clinical information from fecal samples of patients with Behçet's disease uveitis (BU), Vogt-Koyanagi-Harada syndrome (VKH), ankylosing spondylitis (AS), rheumatoid arthritis (RA), Crohn's disease (CD), ulcerative colitis (UC), and healthy controls to form a Behçet's disease uveitis diagnostic dataset;

[0045] 2) KneadData, combined with Trimmomatic and Bowtie2, was used to filter human genes, low-quality sequences and short sequences in the original reads of fecal sample metagenomic sequencing data. The filtered reads were then analyzed for microbial taxonomic composition using MetaPhlan2 to obtain several microbial species taxonomic annotation results and their corresponding gene abundance data. Gene abundance spectrum of Behçet's disease uveitis diagnostic dataset was constructed.

[0046] 3) Using the gene abundance profile obtained in step 2), perform α-diversity analysis and β-diversity analysis to screen out genera and species with statistically significant differences in abundance in uveitis, other immune diseases and healthy controls, and identify specific microbial candidate species that may be associated with BU.

[0047] 4) By using linear discriminant analysis of effect size, identify BU-specific differential species with significant discriminative ability from the "potentially BU-related specific microbial candidate species" screened in step 3);

[0048] 5) Construct ROC curves for the differentially expressed species screened in step 4), select species with AUC > 0.7, and obtain the above biomarker combination.

[0049] Preferably, principal component analysis (PCA) is performed on the Behçet's disease uveitis diagnostic dataset described in step 1). By reducing the dimensionality, the high-dimensional microbial composition data in the Behçet's disease uveitis diagnostic dataset is transformed into visualized principal components, which are used to intuitively display the overall variation of the gut microbiome composition among samples.

[0050] Preferably, a constrained principal coordinate analysis (CPCoA) is performed on the Behçet's disease uveitis diagnostic dataset described in step 1) to determine the differences in gut microbiome composition between BU patients, healthy controls, and patients with other systemic immune diseases in the Behçet's disease uveitis diagnostic dataset.

[0051] Preferably, in step 3), the metagenomic profiling statistical analysis uses the Benjamini-Hochberg method to control the false discovery rate, with a significance threshold of FDR < 0.1.

[0052] The present invention also includes a kit for diagnosing Behçet's disease uveitis (BU), comprising reagents for detecting at least one bacterial species in the above-described combination of biomarkers.

[0053] Preferably, the reagent includes primers, probes, or antibodies targeting strain-specific genes.

[0054] The present invention also includes the use of the above-described biomarker combination in the preparation of reagents or kits for diagnosing Behçet's disease uveitis (BU).

[0055] The present invention also includes the use of the above-mentioned combination of biomarkers in distinguishing patients with Behçet's disease uveitis (BU) from healthy controls, VKH patients, AS patients, RA patients, CD patients, or UC patients.

[0056] The beneficial effects of this invention are as follows:

[0057] 1. By utilizing the 24 specific microbial biomarkers in the biomarker combination described in this invention, a multi-marker combined diagnostic model for diagnosing Behçet's disease uveitis (BU) can be established to achieve high-precision diagnosis of BU, overcoming the limitations of traditional diagnostic models that rely on a single indicator, and significantly improving the accuracy and reliability of diagnosis.

[0058] 2. This invention uses a non-invasive sampling method to measure microbial markers through stool sample testing, avoiding the invasive procedures that may be caused by traditional examination methods, improving patient comfort, and reducing examination costs.

[0059] 3. The 24 microbial biomarkers derived in this invention have good sensitivity and specificity, and can effectively identify BU patients and healthy controls as well as other systemic immune diseases, such as VKH, AS, RA, CD and UC, thus achieving accurate differential diagnosis of a variety of diseases;

[0060] 4. This invention provides new insights into the pathogenesis of bronchopulmonary ulcer (BU) by systematically analyzing the composition and function of gut microbiota, revealing the role of gut microbiota in BU pathogenesis and deepening the understanding of the disease's pathological mechanisms;

[0061] 5. The kit for diagnosing Behçet's disease uveitis (BU) in this invention is easy to operate, requires no complex equipment or professional technicians, and can be widely used in clinical practice. It provides doctors with a reliable auxiliary diagnostic tool, helps improve diagnostic efficiency, reduces the misdiagnosis rate, and ensures that patients receive timely and effective treatment.

[0062] 6. This invention, through a full-chain design of "expanding the control range (excluding interference from other diseases) → data purification (reducing noise error) → statistical error control (reducing random false positives) → targeted screening (preserving BU-specific species) → rigorous verification (ensuring high discriminative power) → combined diagnosis (improving stability)," comprehensively avoids the risk of false positives from the "data source → analysis process → result verification," ultimately achieving high-precision diagnosis of BU. In particular, when distinguishing BU from easily confused diseases such as VKH, AS, and UC, the false positive rate is significantly lower than that of existing technologies.

[0063] Definitions:

[0064] KneadData is a workflow-based tool (essentially a "tool manager") specifically designed for metagenomic data preprocessing. It doesn't directly perform filtering operations; instead, it integrates tools like Trimmomatic and Bowtie2 to automate the process from raw sequencing data to "clean data." In this invention, KneadData (0.6.1) refers to version 0.6.1 of this software.

[0065] Trimmomatic is a tool focused on sequencing data quality control, primarily used for processing raw reads generated by high-throughput sequencing (such as the Illumina platform).

[0066] Bowtie2: A high-efficiency short sequence alignment tool, primarily used to align sequencing reads with a reference genome.

[0067] Alpha diversity: used to measure the diversity of microbial communities within a single sample, reflecting the richness and evenness of species within the sample.

[0068] β diversity: used to measure the differences in microbial communities between different samples, reflecting the degree of variation in community structure between or within sample groups.

[0069] Read length: In this invention, read length refers to the length of a sequencing read, measured in base pairs (bp). During sequencing, some reads that are too short (such as sequences truncated due to quality filtering) may be generated. These short reads typically contain little biological information and can interfere with subsequent analyses (such as genome assembly and species annotation). Therefore, a length threshold (e.g., 80 bp) is usually set to filter out reads shorter than this threshold. For example, "read length < 80 bp" means retaining reads ≥ 80 bp and discarding shorter sequences.

[0070] False positive risk: In medical diagnosis, the false positive risk refers to the probability that an individual who does not have the target disease (such as Behçet's disease uveitis, BU in this invention) will be incorrectly diagnosed with BU when using a certain diagnostic method or biomarker (e.g., healthy individuals or patients with other systemic immune diseases with similar clinical manifestations). False positives can lead to misdiagnosis, not only subjecting individuals to unnecessary medical interventions (such as drug treatment) but also increasing medical costs and delaying targeted treatment for other diseases. Therefore, this is a key issue that this invention aims to avoid.

[0071] Diagnostic efficacy: In the field of medical diagnostics, diagnostic efficacy refers to the ability of a diagnostic method or biomarker to identify a target disease (e.g., Behçet's disease uveitis, BU in this invention) and its practical value. Core measurement dimensions include accuracy, specificity, sensitivity, discriminative power, and reliability. In this invention, "accuracy" refers to the overall probability of correctly distinguishing between "BU patients" and "non-BU individuals (healthy individuals, patients with other similar diseases)"; "specificity" refers to the ability to exclude non-BU individuals (i.e., the probability of not misdiagnosing healthy individuals or patients with other diseases as having BU); "sensitivity" refers to the ability to identify BU patients (i.e., the probability of not missing any BU patients); "discriminative power" refers to the ability to accurately locate BU among "multiple diseases with similar clinical manifestations"; and "reliability" refers to the stability of the results unaffected by external factors such as sequencing platforms and population geographic location. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of the PCA analysis results of gut microbiome composition among different platforms and populations in an embodiment of the present invention;

[0073] Figure 1 (A) is a schematic diagram comparing the gut microbiome composition between two platforms (Illumina HiSeq 2000 and 4000);

[0074] Figure 1 (B) is a schematic diagram comparing the gut microbiota composition among three population groups (China, the Netherlands, and the United States);

[0075] Figure 2 This is a schematic diagram of the CPCoA analysis results of gut microbiota composition in patients with different diseases and healthy controls in an embodiment of the present invention;

[0076] Figure 2 (A) is a schematic diagram of the CPCoA analysis results of gut microbiome composition based on Bray-Curtis distance at the genus level using ANOSIM and healthy controls;

[0077] Figure 2 (B) is a schematic diagram of the CPCoA analysis results of gut microbiome composition based on Bray-Curtis distance at the species level using ANOSIM and healthy controls;

[0078] Figure 3 This is a schematic diagram of the Shannon index analysis results for immune-mediated diseases such as uveitis and healthy controls in an embodiment of the present invention.

[0079] Figure 3 (A) is a schematic diagram comparing microbial diversity at the genus level in uveitis with other immune-mediated diseases and healthy controls.

[0080] Figure 3 (B) is a schematic diagram comparing microbial diversity at the species level in uveitis with other immune-mediated diseases and healthy controls.

[0081] Figure 4 This is a histogram of LDA scores for species with rich differences between the BU group and other combination groups in this embodiment of the invention.

[0082] Figure 5 The ROC curves in this embodiment of the invention are constructed based on the distribution characteristics of 24 characteristic bacterial species in BU, other immune-mediated diseases, and healthy controls.

[0083] Figure 6 This is a schematic diagram of the process for screening microorganisms in a combination of biomarkers according to the present invention. Detailed Implementation

[0084] like Figure 6 As shown, a method for screening microorganisms for the diagnosis of Behçet's disease uveitis (BU) includes the following steps:

[0085] 1) Data download:

[0086] Metagenomic sequencing data and clinical information from fecal samples of patients with Behçet's disease uveitis (BU), Vogt-Koyanagi-Harada syndrome (VKH), ankylosing spondylitis (AS), rheumatoid arthritis (RA), Crohn's disease (CD), ulcerative colitis (UC), and healthy controls were obtained from NCBI SRA to form a Behçet's disease uveitis diagnostic dataset.

[0087] In this embodiment, the data in the Behçet's disease uveitis diagnostic dataset comes from 5 studies in the NCBI SRA database, which obtained metagenomic sequencing data and corresponding clinical information from a total of 695 fecal samples. The study numbers involved are PRJNA 356102, PRJNA 431482, PRJNA 400072, PRJNA 375935, and PRJNA 356225.

[0088] The raw data described above can be converted into a dataset in fastq format using the fast-dump function provided by NCBI SRA. This dataset covers sample information from two sequencing platforms (Illumina HiSeq 2000 and Illumina HiSeq 4000) and three different populations (China, the Netherlands, and the United States), including samples from 24 patients with Behçet's disease and uveitis (BU), 71 patients with Vogt-Koyanagi-Harada syndrome (VKH), 51 patients with ankylosing spondylitis (AS), 95 patients with rheumatoid arthritis (RA), 86 patients with Crohn's disease (CD), 65 patients with ulcerative colitis (UC), and 303 healthy controls.

[0089] To ensure that the sequencing platform and population origin do not significantly affect the subsequent analysis results, this embodiment also requires principal component analysis (PCA) to be performed on the Behçet's disease uveitis diagnostic dataset mentioned in step 1). By reducing the dimensionality, the high-dimensional microbial composition data in the Behçet's disease uveitis diagnostic dataset is transformed into visualized principal components, which are used to intuitively show the overall variation of the gut microbiome composition among samples, in order to determine whether these factors will affect the subsequent analysis and the final conclusion (if factors such as the overall variation of the gut microbiome composition among samples will affect the subsequent analysis and the final conclusion, for example, if obvious separation of the microbiome is found in different platforms or populations, then samples need to be collected again to obtain Behçet's disease uveitis diagnostic data).

[0090] For example, the sample data in the Behçet's disease uveitis diagnostic dataset involves five studies from different sequencing platforms (Illumina HiSeq 2000 and 4000) and different populations (China, the Netherlands, and the United States). PCA analysis was performed on these samples, utilizing scatter plots of the samples in a coordinate system (e.g., Figure 1 As shown in the figure, the clustering patterns of samples from different groups can be observed intuitively. For example, if samples from the same group (such as BU patients) cluster together in the figure, but are clearly separated from other groups (such as healthy controls and VKH patients), it suggests that there are significant differences in the gut microbiota structure among the groups.

[0091] The analysis results showed no significant segregation of the microbiomes from different platforms or populations. This result indicates that the sequencing platform or population origin has minimal interference with the microbiome analysis results and will not significantly affect subsequent analysis results. Therefore, it provides a scientific basis for merging data from multiple sources for unified analysis, ensuring the reliability of subsequent analyses such as biomarker screening.

[0092] Moreover, although PCA is mainly used in this invention to verify the influence of platform and population factors, this presentation of the overall data structure also provides a basic reference for subsequent more in-depth analysis of microbial community structure differences (such as constrained principal coordinate analysis (CPCoA)), helping researchers to grasp the data characteristics as a whole and then carry out subsequent analysis work in a more targeted manner.

[0093] Of particular note is that, in this embodiment, in order to further verify the research hypothesis that "gut microbial dysbiosis is associated with BU and other immune-mediated diseases", a constrained principal coordinate analysis (CPCoA) will be performed on the Behçet's disease uveitis diagnostic dataset described in step 1) to determine the differences in gut microbiome composition between BU patients and healthy controls and other patients with systemic immune diseases in the Behçet's disease uveitis diagnostic dataset.

[0094] This embodiment uses the CPCoA method based on Bray-Curtis distance to analyze the gut microbiota composition of patients with different diseases (including BU, VKH, AS, RA, CD, UC) and healthy controls at both the genus and species levels. Figure 2 (As shown). The results showed that CPCoA at the genus and species levels explained 8.82% of the total variance between groups (P = 1.0 × 10⁻⁶). -4 ) and 3.85% (P = 1.0 × 10 -4 P = 1.0 × 10 -4 This indicates that the differences in gut microbiome composition between groups are statistically significant. In other words, this result directly demonstrates that there are significant differences in the gut microbiome composition between BU patients and healthy controls, as well as patients with other systemic immune diseases, providing a core basis for subsequent screening of BU-specific biomarkers.

[0095] The CPCoA analysis results in this embodiment visually demonstrate the separation trends in the microbial community structure of different groups, further validating the research hypothesis that "gut microbial dysbiosis is associated with BU and other immune-mediated diseases." This significant difference in microbial community structure between groups indicates the existence of specific microbial community characteristics associated with BU disease status, laying the foundation for subsequent analysis (such as LEfSe screening for differentially expressed species) to uncover biomarkers that can be used for BU diagnosis, and also highlighting the feasibility of BU differential diagnosis based on the gut microbiome.

[0096] 2) Data annotation:

[0097] KneadData, combined with Trimmomatic and Bowtie2, was used to filter human genes, low-quality sequences, and short sequences in the raw reads of fecal sample metagenomic sequencing data. The filtered reads were then analyzed for microbial taxonomic composition using MetaPhlan2 to obtain several microbial species taxonomic annotation results and their corresponding gene abundance data. Gene abundance spectrum of Behçet's disease uveitis diagnostic dataset was constructed.

[0098] In this embodiment, KneadData (0.6.1) is used in combination with Trimmomatic (0.39) and Bowtie2 (2.3.5) to filter raw reads (or raw sequencing reads), such as removing human genes and low-quality sequences (i.e., average quality).

[0099] <20) and short sequences (i.e., readings <80bp), and MetaPhlan2 (v2.7.2) was used for microbial taxonomic composition analysis (including bacteria, fungi, eukaryotes and viruses) and gene abundance profile construction;

[0100] MetaPhlan2 (v2.7.2) was used as a dedicated metagenomic classification analysis tool to parse preprocessed sequencing reads, enabling precise classification and identification of microbial communities. This involves accurately identifying various microorganisms, including bacteria, fungi, eukaryotes, and viruses, and clarifying their taxonomic status. For example, the types of microorganisms present in the sample (such as *Bilophila* and *Alistipes*) were identified, and their taxonomic levels (genus, species, etc.) were labeled. This provides fundamental data for subsequent analysis of differences in microbial composition among different groups (such as BU patients, healthy controls, and patients with other immune diseases).

[0101] Under this premise, the relative abundance of different microbial species in the sample can be calculated and a gene abundance spectrum can be constructed, which is the proportion of each species in the sample and is a prerequisite for identifying specific biomarkers.

[0102] Furthermore, this embodiment constructs a gene abundance spectrum using MetaPhlan2 (v2.7.2) to quantify the relative abundance of different microbial species in the sample. In other words, a structured abundance matrix is ​​formed using the relative abundance value of each species in the sample (i.e., the proportion of that species' sequences to the total number of valid sequences in the sample). This abundance information serves as core data support for subsequent diversity analyses (such as α and β diversity analyses), differential species screening (such as LEfSe analysis), and ROC curve validation. For example, only based on the abundance spectrum can the Shannon index be calculated to reflect α diversity, and only then can statistical methods be used to compare the differences in species abundance between groups, thereby screening for species with higher LDA scores and...

[0103] Effective biomarkers with an AUC > 0.7.

[0104] In this embodiment, the final analytical result obtained in this step was "a species-level annotation accuracy of 92.56% ± 6.84% and a strain-level annotation accuracy of 46.27% ± 23.39% in different groups," indicating high annotation efficiency and reliability. This high reliability ensures the scientific validity of subsequent analyses based on microbial composition and abundance (such as CPCoA analysis revealing differences in inter-group microbial community structure and screening 24 characteristic bacterial species), providing a crucial guarantee for the final establishment of a high-diagnostic-efficacy BU biomarker ensemble.

[0105] 3) α and β diversity analysis:

[0106] Following the method described below, α-diversity analysis and β-diversity analysis were performed using the gene abundance profile (including microbial species classification annotation results) obtained in step 2) to explore differences in microbial diversity among groups, obtaining α-diversity difference results, β-diversity analysis results, and a list of species with differential abundance. Genera and species with statistically significant differences in abundance among uveitis, other immune diseases, and healthy controls were screened to identify specific microbial candidate species that may be associated with BU for subsequent screening of specific biomarkers.

[0107] 3-1) First, the data from each sample in the gene abundance spectrum are integrated using the merge_metaphlan_tables.py and metaphlan_to_stamp.pl tools to form a microbial dataset, which is then stored in the computer.

[0108] 3-2) Convert the file format of the microbial dataset to a file format (e.g., SPF format) that can be recognized by metagenomic profiling statistical analysis tools (e.g., STAMP) to obtain an SPF file based on abundance quantitative data, wherein the abundance quantitative data includes the predicted number of reads for each species;

[0109] 3-3) Based on the quantitative data of abundance, the Shannon index at the genus and species levels was calculated using R language, and the differences in microbial richness and evenness at the genus and species levels of different groups (such as BU, VKH, healthy control, etc.) were compared to obtain the α diversity difference results.

[0110] In this embodiment, calculating the Shannon index is the core of alpha diversity analysis, aiming to reveal, as a whole, the differences in microbial richness and evenness within samples among different groups (e.g., BU patients, healthy controls, and patients with other immune diseases). Figure 3 As shown in the figure, this is a preliminary exploration of the "overall diversity characteristics" of microbial communities.

[0111] In other words, this embodiment uses alpha diversity analysis to measure the microbial community diversity within a single sample. The Shannon index is one of the core indicators of alpha diversity, which considers both species richness (number of species) and evenness (distribution uniformity of relative abundance among species) within the community. This step involves calculating the Shannon index at the genus and species levels for different groups (e.g., BU, VKH, AS, RA, CD, UC, and healthy control) and comparing between groups to reveal the differences in microbial diversity within each sample group.

[0112] To construct a more complete gene abundance spectrum from data from different samples, we used Python's `merge_metaphlan_tables.py` and `metaphlan_to_stamp.pl` functions to integrate the sample data generated by MetaPhlan2 and convert them into SPF format files containing predicted read counts. During this process, the relative abundance information of each species in different samples was organized into a unified format, preparing for subsequent calculations and analysis. Then, based on the integrated data, we used tools such as R to calculate the Shannon diversity index at the genus and species levels, further refining the information on species diversity in the gene abundance spectrum, thereby maximizing the construction of a more complete gene abundance spectrum.

[0113] 3-4) Use metagenomic statistical analysis tools (such as STAMP) to perform metagenomic statistical analysis on the SPF files, screen out genera and species whose abundance differs statistically between groups, and form a list of species with differential abundance.

[0114] In this embodiment, the Benjamini-Hochberg method was used to control the false discovery rate in metagenomic profiling statistical analysis, with a significance threshold of FDR < 0.1. In other words, this embodiment utilizes the STAMP tool to analyze SPF files (i.e., uses the STAMP metagenomic profiling statistical analysis tool to compare differences between groups), corrects for multiple tests using the Benjamini-Hochberg method, and screens for genera and species with statistically significant differences in abundance between groups (FDR < 0.1, to identify statistically significant characteristics). Alternatively, the Benjamini-Hochberg method is used to correct for p-values ​​(FDR < 0.1) to clarify the statistical significance of diversity differences.

[0115] 3-5) Extract the abundance data of operational taxonomic units (OTUs) from the SPF file (equivalent to extracting species abundance data from the MetaPhlan2 annotation results) to calculate the distance between samples (e.g., Bray-Curtis distance, i.e., calculate...).

[0116] The Bray-Curtis distance matrix is ​​used, and then PCA / CPCoA plots are drawn using R language (such as the ggplot2 package) to label the distribution of samples from different groups, thereby analyzing the differences in microbial community structure between groups. For example, ANOSIM (to analyze whether the difference between groups is greater than the difference within groups) or PERMANOVA (Permutational Multivariate Analysis of Variance) can be used to verify the significance of differences in community structure between groups (P = 1.0 × 10⁻⁶ for CPCOA analysis). -4 (This indicates that the differences are statistically significant), and the final results of the β-diversity analysis were obtained.

[0117] In other words, in order to compare the microbial community structure of the disease group and the healthy control group, this embodiment extracts operational taxonomic units (OTUs) at different taxonomic levels from the SPF file and uses them to compare the β diversity between different groups (based on the distance between samples with OTU differences in each sample).

[0118] Therefore, this embodiment's screening of specific species (at the genus or species level) with statistically significant differences in abundance between groups represents a "refined differential species mining" based on overall diversity differences. Thus, this embodiment first uses the Shannon index to clarify "overall diversity differences between groups," and then further uses STAMP to screen "which specific species caused this difference," screening and judging microorganisms that can determine BU from a holistic to a local, macroscopic to microscopic perspective.

[0119] In other words, whether it's "calculating the Shannon index using R" or "analyzing SPF files using the STAMP tool," both are based on integrated SPF format files (containing species abundance data for each sample). However, calculating the Shannon index only requires using species abundance data to assess overall diversity and does not involve complex between-group statistical tests. On the other hand, STAMP analysis requires multiple test corrections (such as the Benjamini-Hochberg method) and significance judgment (FDR < 0.1) based on species abundance data, making it a more complex statistical analysis.

[0120] In terms of operational procedures, first completing a simple overall diversity description (Shannon index) and then conducting a complex differential species screening (STAMP) ensures the systematic nature of the screening process and the consistency of the results, providing a complete chain of evidence from "overall characteristics" to "specific species" for subsequent biomarker screening.

[0121] ① α-Diversity differences results: By calculating and comparing the Shannon index at the genus and species levels ( Figure 3 This study aims to clarify the differences in in-sample microbial richness and evenness between BU patients and other patients with immune-mediated diseases (such as VKH, AS, RA, etc.) and healthy controls, and to reveal whether there is an overall imbalance in the gut microbiota of BU patients.

[0122] ② List of species with differential abundance: The SPF files were analyzed using the STAMP tool, and the Benjamini-Hochberg method was used to correct for multiple tests (FDR<0.1) to screen for genera and species with statistically significant differences in abundance among uveitis, other immune diseases and healthy controls, and to identify specific microbial candidate species that may be associated with BU.

[0123] ③ β-diversity analysis results: Based on OTU data extracted from SPF files, the differences in the overall gut microbiota structure between BU patients, healthy controls, and other disease groups were clarified by calculating the distance between samples (reflecting OTU differences), thus verifying the uniqueness of the gut microbiota structure in BU patients.

[0124] These results together constitute the evidence chain that "BU patients have specific gut microbiota dysbiosis", laying the foundation for subsequent screening of highly specific biomarkers (LDA > 3.0) and ROC curve validation (AUC > 0.7) using LEfSe, and ultimately supporting the determination of a combination of 24 BU diagnostic biomarkers.

[0125] 4) Linear discriminant analysis effect size (LEfSe):

[0126] Using linear discriminant analysis of effect size as a biomarker mining method, the "potentially related" biomarkers were selected in step 3).

[0127] From the "BU-related specific microbial candidate species", we can identify BU-specific and significantly discriminative differential species.

[0128] In other words, using the linear discriminant analysis effect size (LEfSe) method, from the "potentially BU-related specific microbial candidate species" screened in step 3), we further mine characteristic microbial biomarkers that can specifically distinguish Behçet's disease uveitis (BU) from other disease groups (such as VKH, AS, RA, CD, UC) and healthy controls, as BU-specific microbial biomarkers with significant discriminative ability. In this embodiment, this mining process is completed through the Galaxy online analysis platform. The α value of the Kruskal-Wallis rank-sum test between classes is set to 0.5 (used to screen species with differences between groups), and the threshold logarithm LDA score of the discriminant feature is 3.0 (e.g., ...). Figure 4 (As shown).

[0129] In other words, this embodiment uses LEfSe to identify BU-specific microbial biomarkers with an LDA > 3.0, yielding BU-specific, significantly discriminative differentially expressed species including microorganisms from the genera *Bilophila*, *Alistipes*, *Dorea*, *Blautia*, *Coprococcus*, *Lachnospiraceae*, *Bacteroides*, *Paraprevotella*, *Streptococcus*, *Eubacterium*, *Oscillibacter*, *Phascolarctobacterium*, *Burkholderiales noname*, *Parasutterella*, *Enterobacter*, and *Escherichia*. These microorganisms are considered BU-specific, significantly discriminative microbial biomarkers. These biomarkers effectively distinguish BU from other groups and are core candidate species for constructing a high-diagnostic-efficacy BU biomarker ensemble.

[0130] 5) Marker verification:

[0131] The differentially expressed species (i.e., core candidate species) selected in step 4) were further validated and optimized in the following manner to finally obtain a specific combination of biomarkers that can be used for the diagnosis of Behçet's disease uveitis (BU):

[0132] 5-1) Data preprocessing:

[0133] Referring to the previous research methods in the literature "A metagenomic study of the gut microbiome in Behcet's disease" (Z. Ye et al.), extremely low abundance groups (species with an average relative abundance of less than 0.001% in any group) were removed to avoid interference from low abundance species that may have no practical diagnostic significance.

[0134] In other words, this embodiment uses the standard of this reference, taking "species with an average relative abundance of less than 0.001% in any group" as the reference for excluding extremely low abundance groups.

[0135] 5-2) Screening for differentially expressed species:

[0136] Based on the LDA scores obtained from the LEfSe analysis in step 4), species with abundance differences between BU and other groups (including 5 immune-mediated diseases and healthy controls) were screened out and used to construct receiver operating characteristic (ROC) curves.

[0137] 5-3) Verify diagnostic efficacy:

[0138] ROC curves were constructed for the selected differentially expressed species, and their ability to distinguish BU from other groups (i.e., BU disease prediction efficacy) was evaluated by calculating the AUC value (area under the curve). Only species with AUC > 0.7 (having good differential diagnostic value) were retained.

[0139] 5-4) Determine the final combination of markers:

[0140] After repeated verification in steps 5-2) to 5-3), the above classifier was used to differentiate BU patients from uveitis types, systemic immune diseases, and healthy controls. The results showed that the AUC of the biomarker composed of 24 bacterial species was 74.83% in BU and VKH, with a 95% confidence interval (CI) of 63.23%-86.43%; the AUC in BU and AS was 98.37%, with a 95% confidence interval (CI) of 96.27%-100.5%; and the AUC in BU and RA was... The AUC was 88.64%, with a 95% confidence interval (CI) of 81.09%–96.20%; the AUC for BU and UC was 99.55%, with a 95% confidence interval (CI) of 98.59%–100.5%; the AUC for BU and CD was 99.03%, with a 95% confidence interval (CI) of 97.78%–100.3%; and the AUC for BU and HC was 82.82%, with a 95% confidence interval (CI) of 72.85%–92.78%.

[0141] The final biomarker combination consisted of 24 species from 16 genera. This combination demonstrated statistical significance in distinguishing BU from VKH, AS, RA, UC, CD, and healthy controls (HC), with AUC values ​​ranging from 74.83% to 99.55%. Figure 5 As shown in the figure, the final combination of biomarkers that can be used for the diagnosis of Behçet's disease and uveitis obtained in this embodiment is as follows:

[0142] The species of microorganisms in the genus Bilophila is Bilophila;

[0143] The species of microorganisms in the genus Alistipes are Alistipesputredinis, Alistipesshahii, and Alistipessp AP11;

[0144] The species of microorganisms in the genus Dorea are Doreaformicigenerans and Dorealongicatena.

[0145] The species of microorganisms in the genus Blautia are Ruminococcus gnavus, Ruminococcus obeum, and Ruminococcus torques;

[0146] The species of microorganisms in the genus Coprococcus is Coprococcus comes;

[0147] The species of microorganism in the genus Lachnospiraceae is Lachnospiraceae bacterium 5_1_63FAA;

[0148] The species of microorganisms in the genus Bacteroides are Bacteroides coprophilus, Bacteroides eggerthii, and Bacteroides plebeius;

[0149] The species of microorganism in the genus Paraprevotella is Paraprevotella;

[0150] The species of microorganisms in the genus Streptococcus are Streptococcus salivarius and Streptococcus thermophilus.

[0151] The species of microorganism in the genus Eubacterium is Eubacterium hallii;

[0152] The species of microorganisms in the genus Oscillibacter is Oscillibacter unclassified.

[0153] The species of microorganisms in the genus *Phascolarctobacterium* are:

[0154] Phascolarctobacteriumsuccinatutens;

[0155] The species of microorganism in the genus Burkholderiales noname is Burkholderiales bacterium1_1_47;

[0156] The species of microorganism in the genus Parasutterella is Parasutterella excrementihominis.

[0157] The species of microorganisms in the genus Enterobacter is Enterobacter cloacae;

[0158] The species of microorganism in the genus Escherichia is Escherichia.

[0159] In summary, through rigorous screening and verification, this embodiment has ultimately determined that the above-mentioned biomarker combination is a BU-specific biomarker combination with high diagnostic efficacy, providing a reliable basis for the early clinical diagnosis of BU and its differential diagnosis with other diseases, and also laying the foundation for the subsequent development of diagnostic kits.

[0160] In other words, the biomarkers selected above possess the unique microbial characteristics of BU, rather than "characteristics common to all immune diseases". They also meet the criteria of "different from healthy individuals" and "significantly different from the other 5 diseases", which can reduce the false positive probability of misdiagnosing other diseases as BU from the source.

[0161] This embodiment also includes the preparation of a kit using the above-mentioned combination of biomarkers. For example, based on the above-mentioned markers, specific primers or probes are designed to construct a real-time PCR kit, which enables diagnosis by detecting the relative abundance of target bacteria in fecal samples.

[0162] It is worth noting that this embodiment uses 24 species from the aforementioned 16 genera for reagent kit preparation, rather than a single species. This means that a single marker may lead to false positives due to "occasional high abundance in a few non-BU individuals," while the combination of multiple markers can reduce this risk through "synergistic discrimination." The "synergistic discrimination model" constructed in this embodiment requires "most markers to conform to BU characteristics" to be considered positive, avoiding false positives caused by the random error of a single marker, and significantly improving the stability and specificity of diagnosis. Experiments have demonstrated that the reagent kit described in this invention can effectively distinguish BU from healthy individuals and diseases such as VKH, AS, RA, CD, and UC.

[0163] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications made to the present invention by those skilled in the art without departing from the spirit of the present invention shall fall within the protection scope of the present invention.

Claims

1. A combination of biomarkers for the diagnosis of Behçet's disease and uveitis, characterized in that, This biomarker set includes microorganisms from the genera *Bilophila*, *Alistipes*, *Dorea*, *Blautia*, *Coprococcus*, *Lachnospiraceae*, *Bacteroides*, *Paraprevotella*, *Streptococcus*, *Eubacterium*, *Oscillibacter*, *Phascolarctobacterium*, *Burkholderiales noname*, *Parasutterella*, *Enterobacter*, and *Escherichia*.

2. The biomarker combination according to claim 1, characterized in that, This biomarker combination includes the following bacterial species: Bilophila, a member of the genus Bilophila; Alistipes genus Alistipesputredinis, Alistipesshahii and Alistipessp AP11; Doreaformicigenerans and Dorealongicatena (genus Dorea); The genus *Blautia* includes *Ruminococcus gnavus*, *Ruminococcus obeum*, and *Ruminococcus torques*. Coprococcus of the genus Coprococcus comes; Lachnospiraceae bacterium 5_1_63FAA of the genus Lachnospiraceae; Bacteroides coprophilus, Bacteroides eggerthii and Bacteroidesplebeius of the genus Bacteroides; Paraprevotella; Paraprevotella; Streptococcus salivarius and Streptococcus thermophilus of the genus Streptococcus; Eubacterium hallii of the genus Eubacterium; Oscillibacter of the genus Oscillibacter unclassified; Phascolarctobacterium succinatutens (a member of the genus Phascolarctobacterium); Burkholderiales bacterium 1_1_47 of the genus Burkholderialesnoname; Parasutterella excrementihominis (genus Parasutterella); Enterobacter cloacae of the genus Enterobacter; Escherichia is a species of Escherichia.

3. A method for screening microorganisms in a combination of biomarkers as described in claim 1, characterized in that, Includes the following steps: 1) Obtain metagenomic sequencing data and clinical information from fecal samples of patients with Behçet's disease uveitis, Vogt-Koyanagi-Harada syndrome, ankylosing spondylitis, rheumatoid arthritis, Crohn's disease, ulcerative colitis, and healthy controls to form a Behçet's disease uveitis diagnostic dataset. 2) KneadData, Trimmomatic, and Bowtie2 were used to filter human genes, low-quality sequences, and short sequences in the original reads of fecal sample metagenomic sequencing data. MetaPhlan2 was used to perform microbial taxonomic composition analysis on the filtered reads to obtain several microbial species taxonomic annotation results and their corresponding gene abundance data, and to construct a gene abundance spectrum for the Behçet's disease uveitis diagnostic dataset. 3) Using the gene abundance profile obtained in step 2), perform α-diversity analysis and β-diversity analysis to screen out genera and species with statistically significant differences in abundance in uveitis, other immune diseases and healthy controls, and identify specific microbial candidate species that may be associated with BU. 4) By using linear discriminant analysis of effect size, we can identify BU-specific differential species with significant discriminative ability from the "potentially BU-related specific microbial candidate species" screened in step 3); 5) Construct ROC curves for the differentially expressed species screened in step 4), select species with AUC > 0.7, and obtain the biomarker combination as described in claim 1.

4. The method according to claim 3, characterized in that, Principal component analysis was performed on the Behçet's disease uveitis diagnostic dataset described in step 1). By reducing the dimensionality, the high-dimensional microbial composition data in the Behçet's disease uveitis diagnostic dataset was transformed into visualized principal components, which were used to intuitively display the overall variation of the gut microbiome composition among samples.

5. The method according to claim 3, characterized in that, A constrained principal coordinate analysis was performed on the Behçet's disease uveitis diagnostic dataset described in step 1) to determine the differences in gut microbiome composition between BU patients, healthy controls, and patients with other systemic immune diseases in the Behçet's disease uveitis diagnostic dataset.

6. The method according to claim 3, characterized in that, In step 3), the metagenomic profiling statistical analysis used the Benjamini-Hochberg method to control the false discovery rate, with a significance threshold of FDR < 0.

1.

7. A kit for diagnosing Behçet's disease uveitis (BU), characterized in that, It includes a reagent for detecting at least one bacterial species in the biomarker combination as described in claim 1.

8. The reagent kit according to claim 7, characterized in that, The reagents include primers, probes, or antibodies targeting strain-specific genes.

9. The use of the biomarker combination as described in claim 1 in the preparation of reagents or kits for diagnosing Behçet's disease uveitis (BU).

10. The use of the combination of biomarkers as described in claim 1 in distinguishing patients with Behçet's disease uveitis (BU) from healthy controls, VKH patients, AS patients, RA patients, CD patients, or UC patients.

Images